RNA processing/modifying enzymes play key roles in the response to thermospermine in Arabidopsis thaliana

SUMMARY

Thermospermine is involved in negative regulation of xylem differentiation by enhancing the translation of mRNAs of the SAC51 gene family in Arabidopsis (Arabidopsis thaliana). These mRNAs contain conserved upstream open reading frames (uORFs) that interfere with the translation of the main ORF. To investigate the mechanism by which thermospermine acts in this process, we isolated mutants insensitive to thermospermine, named ‘its’. We show that the four genes responsible for these mutants, its1 to its4, encode: (i) a homolog of SPOUT RNA methyltransferase, (ii) an rRNA pseudouridine synthase CBF5/NAP57, (iii) a putative spliceosome disassembly factor STIPL1/NTR1, and (iv) a plant‐specific RNA‐binding protein PHIP1. These four mutants were found to have much higher levels of thermospermine than the wild‐type. While all these mutants except its1 appear almost normal, they enhance the dwarf phenotype of a mutant of ACL5, which encodes thermospermine synthase, resulting in tiny plants resembling a double knockout of ACL5 and SACL3, a member of the SAC51 family. Reporter assays revealed that GUS activity from the CaMV 35S promoter‐SAC51 5′‐GUS fusion construct was significantly reduced in its1 and its4 or not affected in its2 and its3, while it was slightly increased in its1, its3, and its4, or not changed in its2 by thermospermine. These findings underscore the critical role of RNA processing and modification in the thermospermine‐dependent translational regulation of uORF‐containing transcripts.

Significance Statement

Thermospermine negatively regulates xylem formation through enhancing mRNA translation of the SAC51 family in Arabidopsis. Isolation and characterization of thermospermine‐insensitive mutants named its reveals the involvement of RNA processing/modification in the action of thermospermine.

Thermospermine functions to repress xylem differentiation by enhancing mRNA translation of the SAC51 family, which is regulated by the upstream open reading frame (uORF) conserved in the 5′ leader of each gene. The identification of four thermospermine‐insensitive mutants and their corresponding genes in Arabidopsis reveals that all of these genes are involved in RNA processing, suggesting its importance in thermospermine‐mediated release of the main ORF translation from the inhibitory effect of the uORF.

INTRODUCTION

Thermospermine is produced from spermidine and an aminopropyl donor, decarboxylated S‐adenosyl methionine, by thermospermine synthase (Knott et al., 2007) and is ubiquitously detected in the plant kingdom (Minguet et al., 2008; Solé‐Gil et al., 2019; Takano et al., 2012). In Arabidopsis thaliana, the ACAULIS5 (ACL5) gene encoding thermospermine synthase is specifically expressed in xylem precursor cells and is required for negative regulation of xylem differentiation (Clay & Nelson, 2005; Kakehi et al., 2008; Muñiz et al., 2008). Loss‐of‐function mutants of ACL5 have an excess of xylem vessels and exhibit a dwarf phenotype (Hanzawa et al., 1997). The phenotype is partially reversed by exogenous thermospermine or a structurally related tetraamine, norspermine (Kakehi et al., 2010). In contrast, the growth of wild‐type seedlings in the presence of thermospermine results in the repression of xylem differentiation and lateral root formation (Tong et al., 2014). Studies of suppressor mutants, which restore the dwarf phenotype of acl5, have identified SUPPRESSOR OF ACL5 (SAC) genes and revealed that thermospermine enhances mRNA translation of SAC51, which encodes a basic helix–loop–helix (bHLH) protein (Imai et al., 2006). The SAC51 mRNA contains upstream open‐reading frames (uORFs) in the 5′ leader sequence, one of which is highly conserved among different plant species and acts to interfere with the main ORF translation (Hayden & Jorgensen, 2007; Jorgensen & Dorantes‐Acosta, 2012; Tran et al., 2008; von Arnim et al., 2014). Mutations in this uORF behave as a dominant SAC, and similar uORF mutations have also been identified in members of the SAC51 gene family, such as SACL1 and SACL3 (Cai et al., 2016; Vera‐Sirera et al., 2015) We further found that other suppressor mutants of acl5, such as sac52‐d, sac53‐d, and sac56‐d, are dominant alleles of RPL10A, RACK1A, and RPL4A, encoding ribosomal protein L10, an integral ribosomal component named Receptor for Activated C Kinase1, and ribosomal protein L4, respectively (Imai et al., 2008; Kakehi et al., 2015). It has been suggested that these three ribosomal gene mutations suppress the acl5 phenotype by reducing the inhibitory effect of the uORFs on the translation of the main ORF in SAC51 family genes.

ACL5 expression is induced by auxin (Hanzawa et al., 2000). Mechanistically, heterodimers of bHLH transcription factors, LONESOME HIGHWAY (LHW) and TARGET OF MONOPTEROS5 (TMO5) or TMO5LIKE1 (T5L1), activate ACL5 expression in xylem precursor cells of the root (Katayama et al., 2015). TMO5 and T5L1 are direct targets of an auxin‐responsive transcription factor ARF5/MONOPTEROS (Donner et al., 2009; Schlereth et al., 2010). LHW‐TMO5 and LHW‐T5L1 heterodimers also activate the expression of SACL3, the cytokinin biosynthetic gene LONELY GUY4 (LOG4), and ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEIN6 (AHP6), which encodes a cytokinin signaling inhibitor (De Rybel et al., 2014; Katayama et al., 2015; Ohashi‐Ito et al., 2014; Vera‐Sirera et al., 2015). SACL3, in turn, antagonizes TMO5 and T5L1, forming a heterodimer with LHW, and consequently functions as a repressor of both ACL5 expression and cytokinin‐dependent provascular cell division (Katayama et al., 2015; Vera‐Sirera et al., 2015). Our previous studies using GUS reporter gene fusions suggested that the translation of the main ORF of SAC51 and SACL1 is significantly enhanced by thermospermine (Cai et al., 2016). In contrast to the overexpression allele sac51‐d, the loss‐of‐function allele sac51‐1 has no additional effect on the acl5 phenotype, whereas that of SACL3, sacl3‐1, severely enhances the acl5 phenotype, resulting in tiny plants. Moreover, the sac51‐1 sacl3‐1 double mutant exhibits an elevated level of thermospermine due to increased expression of ACL5 and shows insensitivity to exogenous thermospermine, with no discernible morphological phenotype (Cai et al., 2016). These results suggest that SAC51 and SACL3 are functionally redundant in the control of ACL5 expression, but SACL3 also plays a role independent of thermospermine.

To further clarify the mode of action of thermospermine, from its perception to the regulation of specific mRNA translation, we screened for thermospermine‐insensitive mutants of Arabidopsis that can grow at high concentrations of thermospermine. Our results unexpectedly reveal that all four genes identified so far, which are responsible for these mutants, encode proteins involved in RNA processing or modification. This provides evidence for the critical roles of RNA metabolism in the plant's response to thermospermine.

RESULTS

Isolation of thermospermine‐insensitive mutants

In the presence of 100 μM thermospermine, wild‐type Arabidopsis seedlings exhibit severely reduced shoot growth, reduced xylem development, and few lateral roots, whereas sac51‐1 sacl3‐1 double mutant seedlings display no reduced growth (Figure 1A). To investigate genetic factors underlying these responses, approximately 10 000 M2 seedlings derived from EMS‐mutagenized seeds were screened for mutants with growth either unaffected or less affected by thermospermine. As a result, 11 mutants were isolated. Genetic mapping and whole‐genome sequencing identified eight alleles across four loci, which were designated as insensitive to thermospermine (its) mutants, its1 including 5 alleles, its2, its3, and its4. Studies on the remaining three mutants are still underway.

Figure 1.

Phenotypes of thermospermine‐insensitive mutants.

(A) Phenotypes of 7‐day‐old seedlings of the wild‐type (Col‐0), acl5, sac51‐1 sacl3‐1, its1‐1, its2, its3, and its4 grown with no (−TS) or 100 μM thermospermine (+TS). Scale bar, 5 mm.

(B–D) Comparison of the length of the first leaf (B), the number of lateral roots (C), and the length of the taproot (D) between Col‐0 and mutant seedlings grown for 7 days on vertically placed MS agar plates with no (−) or 100 μM thermospermine (+). Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

(E) Vascular phenotypes in hypocotyls of Col‐0 and mutant seedlings grown for 7 days with no (−) or 100 μM thermospermine (+). Samples were cleared with chloral hydrate. Scale bar, 200 μm.

(F–H) Comparison of the morphology (F) and the length (G) of the 5th or 6th leaf, and the plant height (H) between Col‐0 and mutant plants grown for 40 days in pots with no thermospermine. Scale bar in (F), 1 cm. In (G) and (H), values are means ± SE (n = 10). Asterisks indicate significant differences from Col‐0 (*P < 0.05; **P < 0.01, Student's t‐test).

The growth of 7‐day‐old seedlings of these mutants is less affected by 100 μM thermospermine than that of wild‐type seedlings (Figure 1A). Measurements of the first foliage leaf length showed that its1, its2, and its4 rather exhibit enhanced growth by thermospermine, whereas its3 is slightly less insensitive (Figure 1B). These mutants also exhibit insensitivity to thermospermine in the number of lateral roots except its1, which restores its reduced number with thermospermine (Figure 1C). However, the taproot length remains unaffected by thermospermine in both wild‐type and mutant seedlings (Figure 1D). Additionally, its1 and its4 seedlings have hypocotyls with a thicker central cylinder, resembling the phenotype of the acl5 mutant (Figure 1E). Unlike the wild‐type or acl5, all its mutants show no or little suppression of the development of the vascular stele by thermospermine (Figure 1E; Figure S1). At the base of the root, the number of xylem vessels is also increased in its1 and its4 7‐day‐old seedlings and not reduced by thermospermine (Figure S1). When grown in pots without thermospermine for 40 days, its1 plants exhibit shorter leaves and reduced height compared with wild‐type plants (Figure 1F–H).

The effect of thermospermine on ectopic xylem differentiation from mesophyll cells in cotyledons was also examined. According to the VISUAL in vitro culture system, Arabidopsis leaf disks or detached cotyledons cultured with auxin, cytokinin, and the brassinosteroid (BR) signaling activator bikinin develop ectopic xylem vessel elements and phloem sieve cells in mesophyll cells (Kondo et al., 2016). In this study, we treated intact seedlings with these compounds. Wild‐type and all its mutants exhibited ectopic xylem vessels around the veins, especially at the tips of cotyledons (Figure 2). Cotreatment with thermospermine narrowed the veins and strongly suppressed ectopic xylem vessel formation in wild‐type and its3 seedlings but had no such effect on its1, its2, and its4 seedlings (Figure 2).

Figure 2.

Effect of thermospermine on ectopic xylem vessel differentiation in cotyledons of Col‐0 and mutant seedlings.

4‐day‐old seedlings grown in agar plates were transferred to liquid media without (Ct) or with auxin, bikinin, and cytokinin (+ABC) or also with thermospermine (+ABCT) for a further 4 days. Samples were cleared with chloral hydrate. Scale bar, 200 μm.

its mutations cause hyperaccumulation of thermospermine

In connection with the thermospermine‐insensitive phenotype, the sac51‐1 sacl3‐1 mutant accumulates higher levels of thermospermine than the wild‐type (Cai et al., 2016). We then examined the level of polyamines in its mutants and found that all four mutants accumulate much higher levels of thermospermine than the wild‐type (Figure 3A). Interestingly, its1 and its4 were shown to accumulate more thermospermine than the sac51‐1 sacl1‐1 sacl2‐1 sacl3‐1 quadruple knockout with levels more than 30 times that of the wild‐type level, while they show reduced levels of putrescine, spermidine, and spermine (Figure 3B–D).

Figure 3.

Polyamine levels and the ACL5 transcript levels in its mutants.

(A–D) Levels of thermospermine (Tspm) (A), putrescine (Put) (B), spermidine (Spd) (C), and spermine (Spm) (D) in wild‐type Col‐0, sac51‐1 sacl1‐1 sacl2‐1 sacl3‐1 quadruple knockout (sacs in short), its1, its2, its3, and its4 mutant seedlings. Polyamines were extracted from 10‐day‐old seedlings grown in MS agar plates. Values are means ± SE (n = 3). Different letters indicate significant differences at P < 0.05 by anova.

(E) The ACL5 transcript levels in Col‐0 and its mutants. Each seedling was grown for 7 days and incubated for 24 h with no (−) or with 100 μM thermospermine (+TS). mRNA levels were normalized to the ACT8 mRNA level and set to 1 in Col‐0 (−). Values are means ± SE (n = 6). Different letters indicate significant differences at P < 0.05 by anova.

Quantitative RT‐PCR analysis revealed that steady‐state ACL5 expression levels are significantly higher in its1, its2, and its4 seedlings than in wild‐type and its3 seedlings. However, after 1‐day treatment with thermospermine, ACL5 expression was downregulated in all the mutants (Figure 3E), indicating that these mutants retain molecular responsiveness to thermospermine.

ITS1 encodes a homolog of SPOUT1

Genome mapping and sequence analysis revealed that five of the total 11 mutants isolated represent different alleles of the gene, At5g19300, which encodes a homolog of the human SPOUT domain‐containing methyltransferase 1 (SPOUT1), also designated as C9orf114. C9orf114 has been reported to interact with mRNA (Baltz et al., 2012; Castello et al., 2012), and it was also shown to bind to a specific miRNA, with its knockdown leading to a reduced level of the mature miRNA, suggesting a role in posttranscriptional regulation (Treiber et al., 2017). The yeast orthologs are thought to play a key role in the formation of the large ribosomal subunit (Ismail et al., 2022). Since At5g19300 has not been characterized previously, we name this locus ITS1 and refer to the five mutant alleles as its1‐1 to its1‐5 (Figure 4A). These alleles contain either a premature termination codon in the protein‐coding sequence or an amino acid substitution in the region conserved among plants and animals (Figure 4B) and behave as recessive mutations (Figure 4C). We further confirmed that a T‐DNA insertion mutant of ITS1, which we name its1‐t, also shows an insensitive phenotype to thermospermine in leaf growth and thick veins in cotyledons (Figure 4C,D).

Figure 4.

Characterization of the gene responsible for its1.

(A) Gene structure of ITS1. Black boxes indicate coding regions and open boxes indicate 5′ leader and 3′ untranslated regions. Mutation sites are indicated by red triangles.

(B) Alignment of amino acid sequences around the mutation sites with the corresponding sequences of the homologous proteins of different organisms. Amino acids substituted in each mutant are highlighted in red. Amino acids identical to those in Arabidopsis are shown in blue. At, Arabidopsis thaliana (NP_197431); Os, Oryza sativa (XP_015637029); Hs, Homo sapiens (XP_047279415); Dm, Drosophila melanogaster (AAM49835).

(C) Comparison of the length of the first leaf between Col‐0 and different its1 seedlings grown for 7 days with 100 μM thermospermine. /+ indicates a heterozygote with the wild‐type allele. Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

(D) Comparison of veins of cotyledons between Col‐0 and its1‐t 7‐day‐old seedlings. Scale bar, 200 μm.

its2 represents a weak allele of CBF5/NAP57

The its2 mutant was found to contain a base substitution in CBF5/NAP57 encoding an rRNA pseudouridine synthase (Figure 5A). In a previous study, two T‐DNA insertion alleles, cbf5‐1 and cbf5‐2, were shown to be lethal (Lermontova et al., 2007). Because the its2 mutant shows recessive inheritance, it was crossed with a heterozygote of another T‐DNA insertion allele, which we refer to here as cbf5‐3 (Figure 5A). The homozygote for this allele was also previously confirmed to be lethal (Kannan et al., 2008). The resulting F1 seedlings showed a 1:1 segregation in terms of sensitivity to thermospermine, and the insensitive seedlings were confirmed to contain the T‐DNA of cbf5‐3 (Figure 5B). Thus, we concluded that its2 is a recessive allele of CBF5/NAP57 and named cbf5‐4. The Gly‐to‐Arg substitution in cbf5‐4 occurs in a Gly residue highly conserved among plants and animals (Figure 5C). However, in contrast to null mutants, cbf5‐4 plants show normal morphology under standard growth conditions, suggesting that it is a weak loss‐of‐function mutant.

Figure 5.

Characterization of the gene responsible for its2.

(A) Gene structure of CBF5/NAP57. A black box indicates a coding region and open boxes indicate 5′ leader and 3′ untranslated regions. Mutation sites are indicated by red triangles.

(B) Comparison of the length of the first leaf between Col‐0 and mutant seedlings grown for 7 days with 100 μM thermospermine. /+ indicates a heterozygote with the wild‐type allele. Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

(C) Alignment of an amino acid sequence around the mutation site of cbf5‐4 with the corresponding sequences of the homologous proteins of different organisms. The amino acid mutated in cbf5‐4 is highlighted in red. Amino acids identical to those in Arabidopsis are shown in blue. At, Arabidopsis thaliana (NP_191274); Os, Oryza sativa (XP_015647556); Mp, Marchantia polymorpha (BBN08575); Hs, Homo sapiens (NP_001354); Dm, Drosophila melanogaster (NP_525120); Sc, Saccharomyces cerevisiae (KZV09420).

(D) Detection of reduced pseudouridylation in the LSU rRNA. Total RNA prepared from Col‐0 and cbf5‐4 seedlings was treated with (+) or without (−) carbodiimide (CMC), subsequently with (+) or without (−) Na₂CO₃ as described in the Methods, and reverse transcribed, followed by PCR amplification of the LSU rRNA sequence containing Ψ sites and agarose gel electrophoresis.

(E) Inverse quantification of pseudouridylation in the LSU rRNA. Total RNA was treated with (+) or without (−) CMC, subsequently with Na₂CO₃, reverse transcribed with rRNA‐specific primers for the LSU rRNA or with an oligo(dT) primer for mRNA, and amplified by PCR. Values are means ± SE (n = 6). Different letters indicate significant differences at P < 0.05 by anova.

We examined whether the level of pseudouridylation in the large subunit (LSU) rRNA is altered in cbf5‐4 by using carbodiimide, which specifically reacts with pseudouridine (Ψ) at alkaline pH, thereby inhibiting reverse transcription of RNAs containing Ψ residues (Motorin et al., 2007). Previous studies have reported Ψ sites in the LSU rRNA in Arabidopsis (Brown et al., 2003; Chen & Wu, 2009; Streit & Schleiff, 2021; Sun et al., 2019). A part of the LSU rRNA containing these Ψ sites (Figure S2) was amplified by RT‐PCR. The results showed that the level of the cDNA fragment amplified from cbf5‐4 was significantly higher than that from the wild‐type, suggesting reduced pseudouridylation in cbf5‐4 (Figure 5D,E). In addition, we examined whether the 5′ leader region of the SACL3 mRNA contains Ψ. The cDNA fragment was amplified to the same degree in both the wild‐type and cbf5‐4, confirming that this region is not pseudouridylated (Figure 5E).

its3 is a semidominant allele of STIPL1 / NTR1

its3 contains a base substitution in SPLICEOSOMAL TIMEKEEPER LOCUS1 (STIPL1) (Figure 6A), which encodes a homolog of the yeast spliceosome disassembly factor, NTC‐Related protein1 (NTR1). However, the region around the amino acid Glu328 replaced by Lys in its3 may not be conserved among plants, animals, and fungi. The rice gene encodes Lys and another Arabidopsis gene, At2g42330, which shares high homology with STIPL1, encodes Arg in this position (Figure 6B). F1 progeny and about half of F2 progeny seedlings from a cross of its3 with the wild‐type show intermediate sensitivity to thermospermine (Figure 6C), indicating that its3 exhibits a semidominant trait. A loss‐of‐function mutant of STIPL1, stipl1‐1, has been identified as a mutant that induces a long circadian period phenotype under constant conditions, with less efficient splicing of circadian‐associated transcripts possibly contributing to the mutant phenotype (Jones et al., 2012). Other studies have reported the involvement of STIPL1 (as NTR1) in a transcription elongation checkpoint at alternative exon (Dolata et al., 2015) and in the promotion of miRNA biogenesis (Wang et al., 2019). A T‐DNA null allele, ntr1‐1 (initially called stipl1‐2), shows pleiotropic phenotypes, including low seed dormancy, delayed flowering, short leaves, and enhanced lethality at high temperatures (Dolata et al., 2015; Wang et al., 2019); however, we detected no altered sensitivity to thermospermine in stipl1‐2 (Figure 6C). To confirm whether its3 is indeed an allele of STIPL1/NTR1, we crossed its3 with stipl1‐2 and found that the F1 seedlings displayed the same insensitivity to thermospermine as the homozygous its3 (Figure 6C). When the T‐DNA construct containing a genomic fragment of the its3 allele of STIPL1/NTR1 with its 1.2‐kb promoter was introduced into stipl1‐2, the resulting transformants recapitulated thermospermine insensitivity in the seedling growth (Figure 6C). We thus concluded that its3 is a semidominant allele of STIPL1/NTR1. Considering another T‐DNA allele ntr1‐2 (Dolata et al., 2015), we now designate its3 as stipl1‐4. In contrast to the temperature‐sensitive stipl1‐2, stipl1‐4 shows seedling growth at 30°C comparable to that of the wild‐type (Figure 6D). However, as with the common phenotype of stipl1‐2, stipl1‐4 exhibits significantly increased and decreased levels of expression of the circadian clock genes LHY and TOC1, respectively, compared with the wild‐type, at 6 h after 10‐day growth under a 16/8‐h light/dark cycle (Figure 6E).

Figure 6.

Characterization of the gene responsible for its3.

(A) Gene structure of STIPL1/NTR1. A black box indicates a coding region, and open boxes indicate 5′ leader and 3′ untranslated regions. Mutation sites are indicated by red triangles.

(B) Alignment of an amino acid sequence around the mutation site of stipl1‐4 with the corresponding sequences of the homologous proteins of different organisms. The amino acid mutated in stipl1‐4 is highlighted in red. Amino acids identical to those in Arabidopsis STIPL1/NTR1 are shown in blue. At, Arabidopsis thaliana (Q9SHG6); Os, Oryza sativa (BAF30296); Mp, Marchantia polymorpha (PTQ30660); Cr, Chlamydomonas reinhardtii (PNW80634); Dm, Drosophila melanogaster (NP_001285636).

(C) Comparison of the length of the first leaf between Col‐0 and mutant seedlings grown for 7 days with 100 μM thermospermine. /+ indicates a heterozygote with the wild‐type allele. stipl1‐2+stipl1‐4 T indicates stipl1‐2 transformed with the T‐DNA containing a full‐length STIPL1 gene of the stipl1‐4 allele. Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

(D) Comparison of the length of the first leaf between Col‐0 and mutant seedlings grown for 2 days at 22°C and further for 2 days at 30°C. Values are means ± SE (n = 10). Asterisks indicate a significant difference from Col‐0 (**P < 0.01, Student's t‐test).

(E) Effect of stipl1‐4 on the expression of clock genes, LHY and TOC1. RNA was prepared from Col‐0 and stipl1‐4 seedlings at 6 h after 10‐day growth under a 16/8‐h light/dark cycle and subjected to qRT‐PCR. mRNA levels were normalized to the ACT8 mRNA level and set to 1 in Col‐0. Values are means ± SE (n = 6). Asterisks indicate a significant difference from Col‐0 (**P < 0.01, Student's t‐test).

its4 is a loss‐of‐function allele of PHIP1

Genome mapping and sequence analysis revealed that its4 is a recessive mutant and contains a base substitution in PHRAGMOPLASTIN‐INTERACTING PROTEIN1 (PHIP1), which encodes a plant‐specific protein with two RNA recognition motifs (Figure 7A). This mutation introduces a premature stop codon in place of Trp271 within a region conserved among many RNA‐binding proteins (Figure 7B). A previous study identified PHIP1 as a binding partner of phragmoplastin, which is implicated in cell plate formation, and suggested a role for PHIP1 in the polarized mRNA transport to the vicinity of the cell plate (Ma et al., 2008). However, no mutants of PHIP1 have been characterized. We tested the sensitivity to thermospermine in a T‐DNA insertion mutant of PHIP1, which we named phip1‐1. The phip1‐1 seedling was moderately insensitive to thermospermine in leaf growth (Figure 7C). We crossed its4, which we now refer to as phip1‐2, with phip1‐1 and found that the insensitivity in F1 seedlings was almost identical to that in homozygous phip1‐2 (Figure 7C). Because the T‐DNA in phip1‐1 is inserted near the end of the protein‐coding sequence, phip1‐1 may be a weak allele. We then generated null mutants using CRISPR gene editing and obtained two alleles with 7‐ and 10‐base deletions in the coding region (Figure 7A). These alleles, named phip1‐3 and phip1‐4, were confirmed to be insensitive to thermospermine (Figure 7C). Based on these results, we concluded that loss‐of‐function mutants of PHIP1 confer a thermospermine‐insensitive phenotype. These phip1 alleles exhibit no morphological abnormalities under standard growth conditions.

Figure 7.

Characterization of the gene responsible for its4.

(A) Gene structure of PHIP1. Black boxes indicate coding regions, and open boxes indicate 5′ leader and 3′ untranslated regions. Mutation sites are indicated by red triangles.

(B) Alignment of an amino acid sequence around the mutation site of phip1‐2 with the corresponding sequences of the homologous proteins of plants and those of animals that have only limited homology in RNA‐binding domains. The amino acid mutated in phip1‐2 is highlighted in red. Amino acids identical to those in Arabidopsis are shown in blue. At, Arabidopsis thaliana (OAP06227); Os, Oryza sativa (EEE62099); Mp, Marchantia polymorpha (PTQ30666); Hs, Homo sapiens (NP_055829); Dm, Drosophila melanogaster (NP_525123).

(C) Comparison of the length of the first leaf between Col‐0 and mutant seedlings grown for 7 days with 100 μM thermospermine. /+ indicates a heterozygote with the wild‐type allele. Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

its mutations affect translation of SAC51 and SACL3

The fact that sac51‐1 sacl3‐1 double knockout mutants are highly insensitive to thermospermine (Figure 1A; Cai et al., 2016) raises a possibility that expressions of SAC51 and SACL3 are affected in its mutants. However, the result of qRT‐PCR experiments revealed that the mRNA levels of all SAC51 family genes are not altered in its mutant seedlings both before and after 24‐h treatment with thermospermine (Figure 8A). We then examined the effect of these its mutations on the mRNA translation of SAC51 by using transgenic lines expressing the SAC51 5′ leader‐GUS fusion gene under the control of the CaMV 35S promoter. The results revealed that the GUS activity was remarkably reduced in its1‐1 and phip1‐2 (its4) seedlings but not in cbf5‐4 (its2) and stipl1‐4 (its3) seedlings (Figure 8B), while the GUS transcript level was not affected significantly in all mutants (Figure S3). We also examined the effect of supplementation of these seedlings with thermospermine on the GUS activity. When grown in the presence of 10 or 50 μM thermospermine, wild‐type seedlings carrying the GUS fusion construct in the wild‐type background had the GUS activity twofold higher than those grown without thermospermine (Figure 8B). The growth with 10 μM thermospermine resulted in a significant increase in the GUS activity compared with that without thermospermine only in stipl1‐4 (its3) and the growth with 50 μM thermospermine increased the GUS activity in its1‐1, stipl1‐4 (its3), and phip1‐2 (its4) but not in cbf5‐4 (its2) (Figure 8B).

Figure 8.

Relationship between its mutations and the SAC51 gene family.

(A) Effect of its1‐1, cbf5‐4, stipl1‐4, and phip1‐2 on the expression of the SAC51 gene family. qRT‐PCR was performed using RNA prepared from Col‐0 and each mutant seedling grown for 7 days and incubated for 24 h with no (−) or with 100 μM thermospermine (+). mRNA levels were normalized to the ACT8 mRNA level and set to 1 in Col‐0. Values are means ± SE (n = 6).

(B) Effect of its1‐1, cbf5‐4, stipl1‐4, and phip1‐2 on the GUS activity generated from the CaMV 35S promoter‐SAC51 5′ leader‐GUS fusion construct. Protein extracts were prepared from 7‐day‐old seedlings of each genotype carrying the GUS construct grown with no (0), 10 μM, or 50 μM thermospermine. The mean GUS activity (n = 10) was normalized against the GUS mRNA level (Figure S3) and shown as the relative ratio to the Col‐0 control without thermospermine. Different letters indicate significant differences at P < 0.05 by anova.

(C) Phenotype of 40‐day‐old plants of acl5 and each double mutant with acl5. Scale bar, 1 cm.

(D) Comparison of the length of the first leaf between Col‐0, sacl3‐d, and each double mutant with sacl3‐d grown for 7 days with no (−) or 50 μM thermospermine (+). Values are means ± SE (n = 10). Different letters indicate significant differences at P < 0.05 by anova.

Unlike sac51‐1, a knockout allele, sacl3‐1 enhances the acl5 phenotype, resulting in tiny‐sized plants (Figure 8C; Cai et al., 2016). To examine whether its mutations also enhance the acl5 phenotype, we crossed each of its mutants with acl5 and found that double mutants of acl5 cbf5‐4, acl5 stipl1‐4, and acl5 phip1‐2 exhibited a tiny plant phenotype like acl5 sacl3‐1, and only acl5 stipl1‐4 plants retained fertility (Figure 8C). So far, we have not obtained acl5 its1 double mutants because the chromosomal locations of ACL5 and ITS1 are very close to each other. We further found that, in contrast to sac51‐1 sacl3‐1, a dominant overexpression allele of SACL3, sacl3‐d, which suppresses the dwarf phenotype of acl5, exhibits an increased sensitivity to thermospermine (Figure 8D). We then examined the thermospermine sensitivity of double mutants of sacl3‐d and each of its alleles. Except for stipl1‐4 (its3) sacl3‐d, the other double mutants exhibited an insensitive phenotype (Figure 8D).

Relationship between its mutations

To get further information on the function of each gene responsible for its mutations, we referred to the co‐expression database ATTED II (Obayashi et al., 2007). The list of the top 50 co‐expressed genes of ITS1, CBF5/NAP57, STIPL1/NTR1, and PHIP1 is shown in Table S1. The gene ontology (GO) analysis of these genes confirmed that ITS1, CBF5/NAP57, and PHIP1 are closely associated with the genes involved in rRNA processing and ribosome biogenesis, while STIPL1/NTR1 is linked with those involved in mRNA splicing (Table S2). In the list of the genes co‐expressed with ITS1, CBF5/NAP57 is ranked as the 17th and, vice versa, ITS1 is the 32nd. CBF5/NAP57 is also ranked as the 43rd in the list of PHIP1 co‐expressed genes. It is noted that more than half of the top 50 co‐expressed genes of CBF5/NAP57 are listed in those of ITS1 or PHIP1.

We next examined expressions of the four responsible genes in each of its mutants. While the ITS1 mRNA level is increased in its1‐1 and cbf5‐4, the CBF5/NAP57 mRNA level is increased only in cbf5‐4 (Figure 9A). The STIPL1/NTR1 mRNA level is decreased in stipl1‐4 and phip1‐2. The PHIP1 mRNA level is also downregulated in phip1‐2 but upregulated in cbf5‐4. Expressions of RPL10A, RACK1A, RPL4A, and JMJ22, identified as a gene responsible for acl5 suppressors, sac52‐d, sac53‐d, sac56‐d, and sac59, respectively, were also examined in its mutants. JMJ22 encodes a D6‐class Jumonji C (JMJD6) protein implicated in RNA processing (Matsuo et al., 2022). We note that JMJ22 is ranked 5th and 47th in the list of co‐expressed genes of ITS1 and CBF5/NAP57, respectively (Table S1). The results revealed that only cbf5‐4 has a slight but significant increase in mRNA levels of RPL10A/SAC52 and JMJ22/SAC59 (Figure 9B).

Figure 9.

Effect of its mutations on their own genes and other SAC genes, and effect of its multiple mutations on the growth.

(A, B) Effect of its1‐1, cbf5‐4, stipl1‐4, and phip1‐2 on the expression of ITS1, CBF5/NAP57, STIPL1/NTR1, and PHIP1 (A) and that of RPL10A/SAC52, RACK1A/SAC53, RPL4A/SAC56, and JMJ22/SAC59 (B). RNA was prepared from 10‐day‐old seedlings of Col‐0 and each mutant and subjected to qRT‐PCR. mRNA levels were normalized to the ACT8 mRNA level and set to 1 in Col‐0. Values are means ± SE (n = 6). Asterisks indicate a significant difference from Col‐0 (*P < 0.05, Student's t‐test).

(C) Comparison of plant height between Col‐0 and mutant plants grown for 30 days in pots. Values are means ± SE (n = 10). Asterisks indicate significant difference from Col‐0 (*P < 0.05, Student's t‐test).

(D) A simplified model illustrating how each of its mutations reduces the sensitivity to thermospermine (334) in the thermospermine‐dependent translation regulation of SAC51 and SACL3. Gray circles indicate small and large ribosome subunits in which Me and Ψ represent methylation and pseudouridine modifications, respectively.

We generated multiple its mutants by crosses. Double mutants of cbf5‐4 stipl1‐4 and stipl1‐4 phip1‐2 are morphologically normal. Triple mutants of its1‐3 cbf5‐4 stipl1‐4 and its1‐3 stipl1‐4 phip1‐2 have no additional phenotype to small leaves and a moderately reduced plant height, which is attributed to its1‐3 (Figure 9C). The cbf5‐4 phip1‐2 double mutants have not so far been obtained because of their close chromosomal locations.

DISCUSSION

In this study, we identified four thermospermine‐insensitive mutants and found that their causative genes encode proteins involved in RNA processing or modification. The phenotypes of these mutants are reminiscent of those observed in sac51‐1 sacl3‐1. Our results showed that GUS reporter activity derived from the SAC51 5′‐GUS fusion is markedly reduced in its1 and its4, while the response to thermospermine is abolished in its2 and attenuated in its3. Double mutants of acl5 with its2, its3, or its4 exhibit a similar tiny plant phenotype, resembling that of acl5 sacl3‐1. These findings suggest that all four its mutants interfere with the action of thermospermine in alleviating the inhibitory effect of the conserved uORF on the main ORF translation of SAC51 and/or SACL3. A simplified model illustrating the role of the four ITS gene products in thermospermine‐dependent translation of the SAC51 family is shown in Figure 9D. Since some response to thermospermine is still retained in these mutants, it is possible that multiple target sites of thermospermine exist, involving different ITS gene products. Moreover, considering that its1 and its4 accumulate higher levels of thermospermine than the quadruple knockout of the SAC51 family, it is also possible that additional regulators of cellular thermospermine homeostasis are affected in these its mutants.

The growth defects observed in knockout alleles of ITS1 indicate the functional importance of ITS1 in other aspects than the response to thermospermine although it is not an essential gene for survival. A recent study has revealed that Upa1 and Upa2, yeast orthologues of SPOUT1/C9orf114, are contained in the primordial pre‐60S ribosomal subunit and raised a possibility that they methylate a specific pseudouridine in the pre‐60S (Ismail et al., 2022). Multiple mutant alleles of ITS1 will provide a tool for biochemical characterization of ITS1 in future research. Although its1 acl5 has not been obtained because of the close chromosomal position of these two mutations, we speculate that the double knockout would display a tiny‐sized plant phenotype with reduced mRNA translation of SACL3.

its2 was found to represent a weak allele of an essential gene CBF5. In other organisms, many studies have suggested the significance of Ψ in RNA‐protein binding (deLorimier et al., 2017; Levi & Arava, 2021; Wu et al., 2016). In yeast, all rRNA Ψ sites are catalyzed by CBF5, and the CBF5‐D95A mutant cells show decreased affinity for tRNAs and decreased translational fidelity (Jack et al., 2011). CBF5/NAP57, which is also called dyskerin, is highly conserved evolutionarily and acts as a core component of a ribonucleoprotein (RNP) complex that associates with RNAs containing the H/ACA motif. Because of the variety of H/ACA RNAs that guide or specify the target of this RNP complex, this protein has been implicated in diverse processes, including ribosome biogenesis, pre‐mRNA splicing by binding to small Cajal body (sca) RNAs, and telomere maintenance by binding to the telomerase RNA (Garus & Autexier, 2021). A recent study has shown that the Arabidopsis CBF5 participates in the plant telomerase complex via interactions with the telomerase RNA despite the lack of a canonical H/ACA motif in the telomerase RNA (Song et al., 2021). Although we have examined and detected the effect of cbf5‐4 only on the pseudouridylation of the rRNA, it is possible that this mutation affects the translation of SAC51, SACL3, and probably other mRNAs through other pathways involving small nucleolar RNAs. We notice that expression levels of the genes including CBF5 itself, ITS1, PHIP1, RPL10A, and JMJ22 are generally upregulated in cbf5‐4, suggesting a feedback control mechanism that compensates for the defect of the CBF5 function.

On the other hand, our results revealed the involvement of a spliceosome disassembly factor STIPL1/NTR1 in mRNA translation of SAC51 and SACL3. The yeast NTR1 gene is essential for cell viability and has been shown to associate with a post‐splicing complex containing the excised intron and the spliceosomal small nuclear (sn) RNAs (Boon et al., 2006). Recently, TFIP11, the human orthologue of NTR1, has been shown to localize to nucleoli and Cajal Bodies and be essential for the 2′‐O‐methylation of U6 snRNA, a core catalytic component of the spliceosome, in the nucleolus (Duchemin et al., 2021). Assembly of U4/U6.U5 tri‐small RNPs is impaired, and the fidelity of pre‐mRNA splicing is affected by the depletion of TFIP11. The Arabidopsis stipl1 knockout mutant shows circadian clock defects. We found that the expression of clock genes is affected in the semidominant stipl1‐4 mutant. However, the mutant of a paralogous gene STIPL2 does not show a circadian phenotype while the double mutants have not been obtained (Jones et al., 2012). The functional difference between these two genes will be elucidated by further investigation with the use of stipl1‐4.

The PHIP1 gene responsible for its4 mutation was initially identified as and named after a phragmoplastin‐interacting protein (Ma et al., 2008). However, no further studies have been reported for PHIP1. The phip1‐2 allele contains a premature stop codon in the middle of two RNA‐binding motifs and may represent a null allele. GO term analysis of co‐expressed genes with PHIP1 strongly suggests the involvement of PHIP1 in ribosome biogenesis and rRNA processing as well as the relation to CBF5. Considering the absence in animals and fungi, it is possible that PHIP1 acts in a plant‐specific RNA modification as‐yet‐unidentified. Although no phenotype is observed in phip1‐2 under standard growth conditions, detailed studies under various growth conditions might uncover the functional importance of PHIP1 other than in the response to thermospermine.

In conclusion, our study reveals that RNA‐modifying and/or processing enzymes play a critical role in the response to thermospermine through activating mRNA translation of the SAC51 family. As we have not yet obtained conclusive data on the effects of each ‘its’ mutation on SAC51 family mRNAs in the polysome profiling experiments, further experiments with improved precision are necessary. In addition, the potential effects of ‘its’ mutations on the translation of other mRNAs containing conserved uORFs remain to be investigated. Given the chemical nature of polyamines, thermospermine may directly interact with RNA molecules. We speculate that the activities of certain RNA‐modifying enzymes are dependent on thermospermine, although the reason is unknown why this specific polyamine is utilized only in plants, particularly in the vascular tissue of vascular plants and in some bacteria. Future studies should focus on the effects of thermospermine on RNA modification, such as those of rRNA and snRNA.

MATERIALS AND METHODS

Plant material and growth conditions

The Arabidopsis (A. thaliana) accession Columbia (Col‐0) was used as the wild‐type. acl5‐1, sac51‐d, sac51‐1, sacl1‐1, sacl2‐1, sacl3‐d, and sacl3‐1 were as described previously (Cai et al., 2016). its mutants were isolated from an ethyl methane sulfonate (EMS)‐mutagenized population of Col‐0 seeds. Briefly, about 5000 seeds of Col‐0 were treated with 0.2% EMS (Sigma, St. Louis, MO, USA) for 16 h, and the M2 progeny seeds were collected as 10 pools of about 500 M1 plants each. About 2000 seeds from each pool were grown on agar plates supplemented with Murashige–Skoog (MS) nutrients (Wako, Tokyo, Japan), 1% sucrose, and 100 μM thermospermine‐4HCl (Santacruz, Dallas, TX, USA) at 24°C under 16‐h light/8‐h dark conditions. Putative mutants were selected visually based on the seedling growth. T‐DNA insertion mutants of ITS1 (SALK_202658C), CBF5/NAP57 (SALK_031065), STIPL1/NTR1 (SALK_073187C), and PHIP1 (SALK_052754) were obtained from the Arabidopsis Biological Resource Center (www.arabidopsis.org). A transgenic line carrying the SAC51 promoter‐5′ leader‐GUS gene fusion construct was as described previously (Ishitsuka et al., 2019). Plants were grown in pots containing rock‐wool cubes surrounded with vermiculite at 22°C under 16‐h light/8‐h dark conditions, except for seedling experiments.

Mapping and genotyping

For initial mapping of its alleles, each mutant was crossed to another wild‐type accession Landsberg erecta (Ler). Genomic DNA was extracted from the F2 seedlings that were grown with thermospermine and showed a thermospermine‐insensitive phenotype. PCR‐based mapping was performed with polymorphic markers as described previously (Konieczny & Ausubel, 1993; Tanaka et al., 2020). Each mutation site was identified by whole‐genome sequencing of the mutant plant DNA using the MGI DNBSEQ‐G400 system at Bioengineering Lab (Sagamihara, Japan).

For generating multiple mutant combinations, genotypes of acl5‐1 and its mutant alleles were confirmed by the dCAPS method (Neff et al., 1998). Genotypes of T‐DNA insertion alleles were confirmed by PCR using respective gene‐ and T‐DNA‐specific primers. Primers and restriction enzymes used are listed in Table S3.

Hormone treatment and microscopy

For ectopic induction of vascular cells in cotyledons, 4‐day‐old seedlings grown on 1/2 MS plates with 1% sucrose were transferred to liquid 1/2 MS media containing 5% glucose, 5 μM 2,4‐D, 1.2 μM kinetin, and 50 μM bikinin with or without 100 μM thermospermine and incubated with gentle shaking for 4 days under 16‐h light/8‐h dark conditions. Seedlings were fixed in a mixture of ethanol and acetic acid (9:1, v/v) for 1 day, cleared with chloral hydrate as described (Yoshimoto et al., 2016), and observed under a light microscope equipped with Nomarski DIC optics (DM5000B, Leica, Wetzlar, Germany).

Gene expression analysis

For RNA preparation, seeds were surface‐sterilized in bleach solution containing 0.01% (v/v) Triton X‐100 for 3 min, washed three times with sterile water, and sown on MS agar plates containing 1% (w/v) sucrose. Seedlings were grown at 24°C under 16 h light/8 h dark conditions. For thermospermine treatment, 10‐day‐old seedlings were preincubated for 24 h in liquid MS media with 1% sucrose and further incubated for 24 h with 100 μM thermospermine. Total RNA was prepared by the phenol extraction procedure (Hanzawa et al., 1997) and reverse transcribed using PrimeScript reverse transcriptase (Takara, Kyoto, Japan). Quantitative real‐time PCR (qRT‐PCR) was performed on three biological replicates using the Thermal Cycler Dice Real Time System TP‐760 (Takara) with the Kapa SYBR fast universal qPCR kit (Kapa Biosystems, Wilmington, MA, USA). ACT8 was used as an internal control. Each sample was duplicated in PCR reactions. Gene‐specific primers used for expression analysis are listed in Table S4.

Detection of RNA pseudouridylation

Ψ in rRNA was detected by using a soluble carbodiimide (Motorin et al., 2007). 10 μg of total RNA was dissolved in 30 μl of the reaction buffer (50 mM Bicine, pH 8.0, 7 m urea, 4 mM EDTA) with 70 mg.ml⁻¹ N‐cyclohexyl‐N′‐β‐(4‐methylmorpholinium) ethylcarbodiimide (CMC) or without CMC as a control and incubated at 37°C for 20 min. The reaction was stopped on ice by addition of 100 μl of 0.3 m sodium acetate at pH 5.2, 700 μl of ethanol, and 1 μl of glycogen at 2 mg.ml⁻¹. After being precipitated at −80°C, RNA pellets were resuspended in 40 μl of 50 mM Na₂CO₃ or in Tris‐EDTA buffer as a control of no alkaline treatment, and incubated at 37°C for 3 h. RNA samples were again precipitated with 1 ml of glycogen solution, 100 μl of 0.3 m sodium acetate, and 700 μl of ethanol at −80°C. Each RNA sample was collected by centrifugation, dried, reverse transcribed with an LSU rRNA‐specific primer, 25S‐R or 25S‐NoPU‐R, or an oligo(dT) primer, and amplified by PCR with specific primers (Table S4). 25S‐NoPU‐R primer was used to amplify the LSU rRNA region that does not contain Ψ sites (Figure S1) as an internal control for qRT‐PCR.

T‐DNA constructs and plant transformation

For recapitulating its3, a genomic fragment harboring the 1.17‐kb promoter and the coding sequence of STIPL1 was amplified from the its3 genomic DNA with gene‐specific primers (Table S5), cloned into a pBI101‐derived Ti plasmid, pTT16 (Takahashi & Komeda, 1989), and introduced into stipl1‐2 through Agrobacterium‐mediated transformation (Clough & Bent, 1998). Transformants were selected by hygromycin resistance.

A CRISPR/Cas9 construct for genome editing of PHIP1 was made by using the pKIR vector (Tsutsui & Higashiyama, 2017) with a pair of oligonucleotides corresponding to the PHIP1 coding sequence (Table S5). The construct was introduced into wild‐type Col‐0. Cas9‐induced mutations were identified by PCR amplification of the target region with gene‐specific primers (Table S5), followed by sequencing.

GUS assay

Fluorometric assay of the GUS activity was performed as described previously (Jefferson et al., 1987). The fluorescence was measured with an RF‐5300PC spectrofluorophotometer (Shimadzu, Kyoto, Japan). Total protein content was measured by using the Bradford assay (Bio‐Rad, Hercules, CA, USA).

Measurement of polyamines

Extraction of polyamines from seedlings, subsequent purification and derivatization were carried out according to Takahashi (2024), and 1,6‐diaminohexane was used as the internal standard. Gas chromatography–mass spectrometry (GC–MS) analysis was adapted from the method of Rambla et al. (2010). A 1‐μl sample of heptafluorobutyrated derivatives dissolved in ethyl acetate was analyzed using the splitless mode of GC–MS (Prominence‐i LC‐2030C, Shimadzu, Kyoto, Japan) with an Rxi‐5Sil MS column (30 m × 0.25 mm × 0.25 μm, Restek, Bellefonte, PA, USA), and fragment ion peaks at m/z 226, 254, 309, 562, and 817 were monitored. The GC program was as follows: 80°C for 1 min, increased to 300°C at 10°C. min⁻¹, and then held for 5 min. Instrument temperatures were set to 200°C for the source, 250°C for the interface, and 250°C for the injector.

Statistical analysis

Data are presented as the mean ± SE for the number of samples indicated in the legends. Statistical significance among the genotypes was evaluated by one‐way analysis of variance (anova) followed by Tukey's test or Student's t‐test.

AUTHOR CONTRIBUTIONS

MS, T. Tanaka, and T. Takahashi designed the experiments; MS, T. Tanaka, and DK performed the experiments; YN and YT measured polyamines; MS, T. Tanaka, YT, HM, and T. Takahashi analyzed the data; MS, T. Tanaka, and T. Takahashi wrote the first draft; and all authors edited the final manuscript.

CONFLICT OF INTEREST

The authors declare that they have no conflicts of interest.

Supporting information

Figure S1. The diameter of the central vasculature in hypocotyls and the number of xylem vessels in roots.

Figure S2. DNA sequence of the LSU rRNA gene.

Figure S3. GUS mRNA levels in its1‐1, cbf5‐4, stipl1‐4, and phip1‐2, carrying the CaMV 35S promoter‐SAC51 5′ leader‐GUS fusion construct.

Table S1. List of the top 50 co‐expressed genes of ITS1, CBF5, STIPL1, and PHIP1.

Table S2. GO analysis of the top 50 co‐expressed genes of ITS1, CBF5, STIPL1, and PHIP1.

Table S3. List of primers for genotyping.

Table S4. List of primers for qRT‐PCR.

Table S5. List of oligonucleotides for T‐DNA construction and edited mutant detection.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.